brad@looking.UUCP (Brad Templeton) (05/02/88)
In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Richard Segal) writes: >One problem with the problem of (F)ollowingup vs. (R)eplying... >As I think we all know by now, (R)eply NEVER works. The variety of mailers >is such that it is impossible to hit (R) and place decent odd on the msg's >arriving. Sadly, this hits the nail on the head, at least of late. If we could actually work out the cost to us in terms of: a) Long signatures made by people who don't trust mailers b) The cost of long bounced messages c) The cost of postings that say, "I couldn't get mail through to Joe, so here's a private message for him" in terms of transmission and wasted human reading time d) The cost of followups that would have been mail if it worked more reliably That we would find the horrid broken mailer problem is actually costing a great deal. We all sit around, not having the time to fix things, but that's af false economy. We can fix this, though. First we have to insist (rather than suggest) that any site that wants a news or mail feed that shows to the outside world *HAS* to register itself with the maps. How do we insist on this? The backbone sites modify their inews to examine the "From" line of every message. If the domain there isn't found in the database of known domains, the message is tossed on the floor. What about the time between hook up and distribution of your registration? There are a couple of answers: a) You don't get to post for the first while, just listen b) New site registrations get updated daily to backbone sites, or c) For the first while you generate a From address using your feed's name, such as user%mysite@feed.uucp, and then switch to a proper address when your name is registered. I'm tired of watching up to 30% of mail bounce, and I'm one hop from a backbone site. We can do this, and it only requires a fairly simple change (in the case of option a) on a couple of dozen machines. We can do this NOW. -- Brad Templeton, Looking Glass Software Ltd. - Waterloo, Ontario 519/884-7473
karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) (05/03/88)
>As I think we all know by now, (R)eply NEVER works. The variety of mailers >is such that it is impossible to hit (R) and place decent odd on the msg's >arriving. Sadly, this hits the nail on the head, at least of late. Speak for yourself, please. My mail gets where it's going well over 90% of the time. I have written 26 pieces of mail today, 13 of them to off-site addresses. None have bounced. Enough of it has already gotten where it's going (including via UUCP transport) that I have exchanged full conversations with 3 individuals, collectively being 9 of the 13 off-site pieces I wrote. --Karl
karl@triceratops.cis.ohio-state.edu (Karl Kleinpaste) (05/03/88)
How do we insist on this? The backbone sites modify their inews to examine the "From" line of every message. If the domain there isn't found in the database of known domains, the message is tossed on the floor. [1] Just how much CPU do you think I've got? I'll grant you that I already have to do a DBM lookup on the Message-ID of every article that comes in, but doubling the number of large lookups of that type does not make me feel good. I have a dual-processor Pyramid 98x whose load is currently at about a 12, and he's just not happy right now. Work is getting done, but he's not happy. [2] Your scheme obliterates 9 of my 23 news neighbors. They won't like that, and I don't like it, notwithstanding the fact that I think they ought to have registered by now. --Karl
friedl@vsi.UUCP (Stephen J. Friedl) (05/03/88)
In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Richard Segal) writes: >One problem with the problem of (F)ollowingup vs. (R)eplying... >As I think we all know by now, (R)eply NEVER works. The variety of mailers >is such that it is impossible to hit (R) and place decent odd on the msg's >arriving. I appreciate those not cluttering the net with maps of the world or with the collected works of Shakespeare, but *no* .sig makes mail rough. I am slowly getting the hang of the various mailer systems, but I hate to try and try and try and try to send mail to somebody via Path and not be able to do it :-(. This goes for private mail not sent from the newsreader as well; sometimes I can't respond. If you don't have a .signature, could you please make a small one with your name and address? Thanks! -- Steve Friedl V-Systems, Inc. (714) 545-6442 3B2-kind-of-guy friedl@vsi.com {backbones}!vsi.com!friedl attmail!vsi!friedl
brad@looking.UUCP (Brad Templeton) (05/03/88)
CPU overhead is not a problem in this. Even leaf sites do the message-id lookup. If you keep message-ids for 1 month, that's around 50,000 entries. There might be only 5 to 10,000 entries in the mail routing database. Most sites look up every mail message in the database, and backbone sites already handle more mail than news. It's a hash lookup, after all. Quite fast. And you only have to keep the domains. For example ".att" ".arpa" ".bitnet" ".edu" ".com" etc. just need that one entry for all sites underneath them, since these are maintained domains with maintained mailers. And we only have to do this at enough sites to get the message through: Get your mailer in gear, or get out. I am glad some people are facing reliable mail. But I know, based on what I see in net traffic, that a lot of people, myself included, are not. Somebody saying, "gee, my mail usually makes it" is not a reason to do nothing. -- Brad Templeton, Looking Glass Software Ltd. - Waterloo, Ontario 519/884-7473
chip@ateng.UUCP (Chip Salzenberg) (05/04/88)
In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Richard Segal) writes: >One problem with the problem of (F)ollowingup vs. (R)eplying... >As I think we all know by now, (R)eply NEVER works. The variety of mailers >is such that it is impossible to hit (R) and place decent odd on the msg's >arriving. I get at least 50% success on replies. Part of it is examining the message before I send it. Sometimes I get stupidities like: To: foobar@xyzzy.UNM.EDU.UUCP which can be fixed before you mail them. If you can't R)eply, complain to your postmaster! Making it work (mostly!) is his job. -- Chip Salzenberg "chip@ateng.UU.NET" or "codas!ateng!chip" A T Engineering My employer may or may not agree with me. "I must create a system or be enslaved by another man's." -- Blake
david@ms.uky.edu (David Herron -- One of the vertebrae) (05/04/88)
In article <1598@looking.UUCP> brad@looking.UUCP (Brad Templeton) writes: >It's a hash lookup, after all. Quite fast. And you only have to keep the >domains. Hmmm... pre-supposing that hash lookups exist everywhere. Let me remind you that dbm doesn't exist on SysV. As for keeping only the domains? well, you're wanting us to check into our database for hostnames to check validity. How far down the domain name should we check? The first level? second level? What happens with naive installers of rn at a domain'd site? "rn" assumes a top-level domain name of "uucp" and sticks the output of /bin/hostname before that. This can easily become a From: david@ms.uky.edu.UUCP hmmm Then there's the rn installation at either lanl or lll which is generating Reply-to: user@a.UUCP I don't know if they're still doing that, the last time I remember seeing it was some months ago but the first time was a year before that. hmmm... Oh, it was "machine a" at some place. Also the style of how to do the lookup is different on different systems. Here I'd like to link in substantial parts of mmdf into news, this would give me access to a sizeable dbm based database plus access to the nameservers. But if this is to be done I'll probably have to do it myself. >For example ".att" ".arpa" ".bitnet" ".edu" ".com" etc. just need >that one entry for all sites underneath them, since these are maintained >domains with maintained mailers. Hah! You've obviously not ever lived on bitnet! While the names are controlled, the maintainence of mailers or even existance of mailers is a very questionable idea. There is a table maintained by the BITNIC which lists all the sites and various details about how to send mail to the site. Like, you send it with BSMTP to some MAILER@HOST. Or you punch it directly to the user. Or any number of possibilities. That's a lot of gaul to begin with to assume that every site knows all the different variations of sending mail. To make it worse the file is in some strange IBM format that I (as a Unix guru) know nothing about. This site is unable to make use of the information and thus isn't able to send mail into bitnet very well. Other sites must be in the same boat. We've been advertising in that file that we can receive BSMTP for 1 1/2 yrs now. We still get mail here that is directly delivered to the user bypassing the mailer. Sorry, I tend to rave about bitnet ... >I am glad some people are facing reliable mail. But I know, based on what >I see in net traffic, that a lot of people, myself included, are not. >Somebody saying, "gee, my mail usually makes it" is not a reason to do >nothing. I agree. On the other hand, "my mail makes it 90% of the time". -- <---- David Herron -- The E-Mail guy <david@ms.uky.edu> <---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET <---- Windowing... The Amiga has windowing. The Mac has windowing (echoes of <---- Jonathan Livingston Seagull: "Just flying? A mosquito can do that much!").
mlandau@bbn.com (Matt Landau) (05/07/88)
In news.misc (<259@ateng.UUCP>), chip@ateng.UUCP (Chip Salzenberg) writes: >I get at least 50% success on replies. Part of it is examining the message >before I send it. Sometimes I get stupidities like: > > To: foobar@xyzzy.UNM.EDU.UUCP > >If you can't R)eply, complain to your postmaster! Making it work (mostly!) >is his job. Actually, in a case like this, I'd say "Complain to usenet@xyzzy.unm.edu" and tell him or her that the local news system is generating bogus headers. There are a tremendous number of sites that just can't seem to get their acts together well enough to generate sensible From: lines in the first place. Rather than try to get the replying site to compensate for brain damaged senders, let's all put pressure on the senders make sure their news installations are doing things correctly. Of course, this still doesn't deal with things like mixed-mode addresses in the From: or Path: lines or mailers that don't grok internet style addresses or uucp-style addresses. THESE are issues to discuss with your local Postmaster -- if you can't reply to a properly-formed From: line, then your mail system is arguably broken. -- Matt Landau Let not a man glory in this: that he loves his country. mlandau@bbn.com Let him glory rather in this: that he loves his kind.
jailbait@dasys1.UUCP (Richard Segal) (05/09/88)
In article <259@ateng.UUCP> chip@ateng.UUCP (Chip Salzenberg) writes: >In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Me, of all people) writes: >>As I think we all know by now, (R)eply NEVER works. >I get at least 50% success on replies. Part of it is examining the message >before I send it. Sometimes I get stupidities like: > > To: foobar@xyzzy.UNM.EDU.UUCP > >which can be fixed before you mail them. > >If you can't R)eply, complain to your postmaster! Making it work (mostly!) >is his job. >-- >Chip Salzenberg "chip@ateng.UU.NET" or "codas!ateng!chip" Well, for me, the biggest problem comes in when the address in the article is a UUCP address, so the rn mailer auto-paths it differently than it would should I put the address down as Foobar@faunch.uucp in the to: line. If the address starts out UUCP then I get 20 or so site path that almost always HAS to be corrected. With the UUCP address defined by me, the pathing AFTER I mail the item usually works. (Did I just make any sense?) By the way, I got one note (From Eric Raymond) noting that he had no problems with reply when I had just gotten a <R>eply I had sent him from Soc.Net-People bounced back to me as undeliverable. I think my reply from within the mailer (as opposed to RN) got to him. *Sigh*, Slack, Jailbait. -- Richard Segal NYU Student, Rev., Looney at Large Big Electric Cat Public UNIX ARPA:SEGAL@ACFCluster.NYU.EDU ..!cmcl2!phri!dasys1!jailbait BITNET:SEGAL@NYUACF.bitnet "Bob", Eris, Norton I, Mal-2, Jailbait. Doesn't it have a nice ring to it?
jv@mhres.mh.nl (Johan Vromans) (05/09/88)
From article <259@ateng.UUCP>, by chip@ateng.UUCP (Chip Salzenberg): > If you can't R)eply, complain to your postmaster! Making it work (mostly!) > is his job. Good grief! R)eply usually works if the article contains a valid address. But lots of sites use <whatever-they-are-used-to>.UUCP, which causes most of the failures. The only thing I can do about that, is to program my mail system to not accept ".UUCP" addresses. BTW, my R)eply score is about 80-90%. -- Johan Vromans | jv@mh.nl via European backbone Multihouse N.V., Gouda, the Netherlands | uucp: ..{uunet!}mcvax!mh.nl!jv "It is better to light a candle than to curse the darkness"
mike@turing.UNM.EDU (Michael I. Bushnell) (05/10/88)
In article <10944@jade.BBN.COM> mlandau@bbn.com (Matt Landau) writes: >In news.misc (<259@ateng.UUCP>), chip@ateng.UUCP (Chip Salzenberg) writes: >>I get at least 50% success on replies. Part of it is examining the message >>before I send it. Sometimes I get stupidities like: >> >> To: foobar@xyzzy.UNM.EDU.UUCP >> >>If you can't R)eply, complain to your postmaster! Making it work (mostly!) >>is his job. > >Actually, in a case like this, I'd say "Complain to usenet@xyzzy.unm.edu" >and tell him or her that the local news system is generating bogus headers. >There are a tremendous number of sites that just can't seem to get their >acts together well enough to generate sensible From: lines in the first >place. Rather than try to get the replying site to compensate for brain >damaged senders, let's all put pressure on the senders make sure their >news installations are doing things correctly. Sheesh. OK, fine. Now that our local stupidity is being discussed on usenet, I'll fix it NOW. Time for that rn reconfig... Actually, though, this is rue. When our site generates bad headers, and it causes problems, tell me and I'll fix it. When the header is correct, tell you local postmaster that it can't be delivered. The real problem, though, is that sometimes people don't know whether the address is correct. The proper solution, in such cases, is to talk to your local net.guru, and have him complain to the remote site if applicable. -- N u m q u a m G l o r i a D e o Michael I. Bushnell HASA - "A" division 14308 Skyline Rd NE Computer Science Dept. Albuquerque, NM 87123 OR Farris Engineering Ctr. OR University of New Mexico mike@turing.unm.edu Albuquerque, NM 87131 {ucbvax,gatech}!unmvax!turing.unm.edu!mike
fair@ucbarpa.Berkeley.EDU (Erik E. Fair) (05/11/88)
A most interesting point of view, Brad. I submit, however, that what we really need to do is bundle a replacement for the ancient /bin/mail program in (V7, sIII, sV.*) along with the netnews distribution. /bin/mail has not been adequate to the needs of any networked organization for a number of years, and as long as AT&T continues to pretend that it is, we have to fend for ourselves. Anybody have a good, public domain replacement that is easily configured, can deal with domains, has mail alias support, can deliver local mail with the appropriate mailbox locking, and can read a pathalias generated path list to get out to the UUCP network? Erik E. Fair ucbvax!fair fair@ucbarpa.berkeley.edu
wisner@eddie.MIT.EDU (Bill Wisner) (05/12/88)
Erik Fair: >Anybody have a good, public domain replacement that is easily >configured, can deal with domains, has mail alias support, can >deliver local mail with the appropriate mailbox locking, and can >read a pathalias generated path list to get out to the UUCP network? Ever heard of smail? ..b
heiby@mcdchg.UUCP (Ron Heiby) (05/13/88)
Erik E. Fair (fair@ucbarpa.Berkeley.EDU) writes: > Anybody have a good, public domain replacement that is easily > configured, can deal with domains, has mail alias support, can > deliver local mail with the appropriate mailbox locking, and can > read a pathalias generated path list to get out to the UUCP network? Ever hear of "smail"? I've been using it for a couple of years. Now at version 2.5, it is available from the various comp.sources.unix archives. I am running on a fairly vanilla SVR3 machine. I found it easy to configure. It can deal with domain addresses. It has mail alias support. It uses the original /bin/mail to deliver local mail. It can read a pathalias generated path list to determine "optimal" routing. The current version does not directly support alternate mailers (as far as I can tell). All my mail is either local delivery or gets passed to uux. There is information in the installation docs for installing it in a "sendmail" environment. I did that once. I found that sendmail wasn't buying me anything. My LAN understands uux/uucico transfers just fine. -- Ron Heiby, heiby@mcdchg.UUCP Moderator: comp.newprod & comp.unix "I believe in the Tooth Fairy." "I believe in Santa Claus." "I believe in the future of the Space Program."
mouse@mcgill-vision.UUCP (der Mouse) (05/16/88)
In article <9164@e.ms.uky.edu>, david@ms.uky.edu (David Herron -- One of the vertebrae) writes: > In article <1598@looking.UUCP> brad@looking.UUCP (Brad Templeton) writes: >> It's a hash lookup, after all. Quite fast. > Hmmm... pre-supposing that hash lookups exist everywhere. Let me > remind you that dbm doesn't exist on SysV. Why not? I see no excuse for this. I just skimmed the source, and I expect it would take me no more than an afternoon to make it work on SV, most of that because of my unfamiliarity with SV. I am certain it could be rewritten to work there. Can someone who knows tell us why dbm or equivalent doesn't exist for SV? Seems to me it isn't much of a "standard" if it doesn't even have a database library. (But then, true Berkloids have always known that. "From now on, consider it sub-standard.") der Mouse uucp: mouse@mcgill-vision.uucp arpa: mouse@larry.mcrcim.mcgill.edu
mouse@mcgill-vision.UUCP (der Mouse) (05/16/88)
In article <4325@dasys1.UUCP>, jailbait@dasys1.UUCP (Richard Segal) writes: > In article <259@ateng.UUCP> chip@ateng.UUCP (Chip Salzenberg) writes: >> In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Me, of all people) writes: >>> As I think we all know by now, (R)eply NEVER works. Not so, not at all. It works for me. At least 90% of the time. >> Sometimes I get stupidities like: >> To: foobar@xyzzy.UNM.EDU.UUCP If you see a domainoid address with .UUCP on the end, complain to usenet at that site. This sort of thing is symptomatic of mis-installed netnews software. > Well, for me, the biggest problem comes in when the address in the > article is a UUCP address, so the rn mailer auto-paths it differently > than it would should I put the address down as Foobar@faunch.uucp in > the to: line. If the address starts out UUCP then I get 20 or so > site path that almost always HAS to be corrected. With the UUCP > address defined by me, the pathing AFTER I mail the item usually > works. This sounds as though your newsreader is configured to use the Path: header line for replies. This is very, very wrong. At best it is horribly inefficient. Usually, there will be at least one hop in there that is a news feed but not a mail feed; if you're lucky, the "sending" machine will route the mail for you. (My opinion is that this option (use the Path: line for replies) should never have existed. The Path: line is not, and never has been, intended as a reply path, so nobody ensures that it's valid as a reply path. Consequently, it often (usually?) isn't.) Use Reply-To:, if it exists; otherwise use From:. der Mouse uucp: mouse@mcgill-vision.uucp arpa: mouse@larry.mcrcim.mcgill.edu
david@ms.uky.edu (David Herron -- One of the vertebrae) (05/17/88)
In article <1104@mcgill-vision.UUCP> mouse@mcgill-vision.UUCP (der Mouse) writes: >In article <9164@e.ms.uky.edu>, david@ms.uky.edu (David Herron -- One of the vertebrae) writes: >> In article <1598@looking.UUCP> brad@looking.UUCP (Brad Templeton) writes: >>> It's a hash lookup, after all. Quite fast. >> Hmmm... pre-supposing that hash lookups exist everywhere. Let me >> remind you that dbm doesn't exist on SysV. > >Why not? I see no excuse for this. > >I just skimmed the source, and I expect it would take me no more than >an afternoon to make it work on SV, most of that because of my >unfamiliarity with SV. I am certain it could be rewritten to work >there. You're right about it being run-able on SysV ... I have it running on a couple of SysV systems here, no changes to the source at all. It doesn't exist on SysV because it doesn't come as part of the standard distribution. I think it didn't come into being until after V7 ... SysV is derived from V6 -> PUB/Unix -> SysIII, and lacks a number of things added in V7. But enough history. There's those PD DBM routines that are in (I think) alt.sources, but I haven't looked at them yet. -- <---- David Herron -- The E-Mail guy <david@ms.uky.edu> <---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET <---- <---- Goodbye RAH.
dg@lakart.UUCP (David Goodenough) (05/18/88)
From article <1105@mcgill-vision.UUCP>, by mouse@mcgill-vision.UUCP (der Mouse): > In article <4325@dasys1.UUCP>, jailbait@dasys1.UUCP (Richard Segal) writes: >> In article <259@ateng.UUCP> chip@ateng.UUCP (Chip Salzenberg) writes: >>> In article <4203@dasys1.UUCP> jailbait@dasys1.UUCP (Me, of all people) writes: >>>> As I think we all know by now, (R)eply NEVER works. > > Not so, not at all. It works for me. At least 90% of the time. > >>> Sometimes I get stupidities like: >>> To: foobar@xyzzy.UNM.EDU.UUCP Much other stuff deleted. There is always the vn approach - it prints a list of all likely looking addresses it can find in the article, and you get to play multiple choice, OR you just type in by hand where you want it to go. I've never had a problem with it (plug plug for vn :-) :-) -- dg@lakart.UUCP - David Goodenough +---+ | +-+-+ ....... !harvard!adelie!cfisun!lakart!dg +-+-+ | +---+